# iOS run 2026-10-25 on EAS (/docs/bench/ios-2026-10-25) The run behind the 2026-10-25 iOS dataset on the [comparison page](/docs/bench/comparison). Everything below comes from the four workers' artifacts, merged into one report; nothing was measured on a developer machine. ## How it was run [#how-it-was-run] | | | | ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | Workflow | `apps/example/.eas/workflows/bench-baseline.yml`, EAS Workflows run [`01a172f5-fdee-7992-b66c-a2173338eb32`](https://expo.dev/accounts/delacour/projects/warden/workflows/01a172f5-fdee-7992-b66c-a2173338eb32), inputs `runs=5`, `frameworks=warden,argent,agent-device,tester-army`, `max_minutes=55` | | Code | warden commit `5ffd64f2` (branch `feature/devboxes`), every suite under `packages/e2e-bench/suites/` | | Workers | One EAS `macos-large` worker per framework, four in all, two at a time on this plan. VM template `macos-tahoe-26.5-xcode-26.6`: VirtualMac2,1, Apple M4 Pro (Virtual), 10 cores, 40 GB, macOS 26.5.2 (25F84), Xcode 26.6 (17F113) | | Simulator | Each worker claims its own headless simulator with `warden claim ios --headless` (iPhone 17, iOS 26.5, named `warden-iphone-17-1`), created fresh for the job and shut down at the end. Nothing one framework leaves behind can reach another | | App | The same build on every worker: `nz.co.delacour.warden.example`, EAS `bench` profile (Release, simulator), sha256 `2d35643eda49`, 101.1 MB, reused by fingerprint (`build_ios` skipped), installed with `simctl install` before the bench | | Runs | 5 runs of the 8-test suite per framework. Run 1 is cold; warm numbers are the median of runs 2 to 5. `retries=0`, fresh app launch per test, `caffeinate -dims` held | | Before the bench | `bun install`, `warden-simd` built from source (Swift release), the framework installed (`bun add -g @swmansion/argent@0.30.1` / `agent-device@0.21.31`; tester.army's `e2e` 0.20.0 comes with `bun install`, Node 22.23.1), then the job waits for the 1-minute load average to drop under the core count and records a 60 s idle control of the simulator with the app in the foreground | | Deep-link sheet | iOS shows "Open in …?" for the first custom-scheme link after a boot and never again, so the harness opens one link and answers the sheet before every framework's prepare. No suite handles the sheet itself | | Hard stop | A watchdog stops the bench at `max_minutes`; it did not fire on any worker (the longest job ran 28 min) | | Isolation per framework | Helpers of every framework stopped, driver apps uninstalled before the run (cold cost measured), residue listed after teardown | | Sampling | Host and simulator processes sampled every 500 ms from `ps`; the CPU headline is CPU-seconds of the framework's own process tree during the timed run, warm median | | Evidence on exit | Each job screenshots the simulator and lists the apps still running, releases the lease even when a step failed, and uploads an `eas-ios-` artifact | | Versions | bun 1.3.14, warden 0.1.1, Argent 0.30.1, agent-device 0.21.31, tester.army e2e 0.20.0 (bundled agent-device 0.21.25), Node 22.23.1 | Hosted workers idle hotter than a Mac and share their host, so these numbers compare only with other `macos-large` runs, never with runs on other hardware. ## Results [#results] ### Summary [#summary] | Framework | Status | Version | Suite median (warm) | p95 | Ratio | Cold rep 1 | Cold-start cost | Prepare | Teardown | Overhead | Pass rate | Flake | Residue | Footprint (LOC / deps) | | ------------ | ------ | ----------------------- | ------------------- | -------- | ----- | ---------- | --------------- | ------- | -------- | -------- | --------- | ----- | ----------------------- | ---------------------- | | warden | ok | bun 1.3.14 + warden-e2e | 97.04 s | 98.63 s | 1.06× | 98.33 s | 1.94 s | 652 ms | 85 ms | 80 ms | 100% | 0% | 0 proc / 0 app / 0 path | 379 LOC / 0 B | | argent | ok | 0.30.1 | 131.81 s | 132.14 s | 1.44× | 132.43 s | 2.35 s | 1.73 s | 511 ms | 168 ms | 100% | 0% | 0 proc / 0 app / 1 path | 161 LOC / 0 B | | agent-device | ok | agent-device 0.21.31 | **91.77 s** | 91.93 s | 1.00× | 92.86 s | 18.41 s | 17.32 s | 322 ms | 223 ms | 100% | 0% | 0 proc / 1 app / 1 path | 158 LOC / 0 B | | tester-army | ok | e2e 0.20.0 | 132.34 s | 133.43 s | 1.44× | 151.03 s | 18.69 s | 0 ms | 523 ms | 2.21 s | 100% | 0% | 0 proc / 1 app / 1 path | 307 LOC / 0 B | ### Per-test medians (ms) [#per-test-medians-ms] | Test | warden | argent | agent-device | tester-army | | ---------------------------------------------------------- | ----------------- | ------------- | ----------------- | ------------- | | onboarding steps through to sign-in | **3136 (1.00×)** | 8220 (2.62×) | 3867 (1.23×) | 6032 (1.92×) | | invalid email and short password show field errors | 17987 (1.09×) | 17962 (1.09×) | **16549 (1.00×)** | 20867 (1.26×) | | a locked account shows the form alert | 11046 (1.14×) | 14040 (1.45×) | **9671 (1.00×)** | 12799 (1.32×) | | valid credentials land on todos | 10937 (1.12×) | 14982 (1.53×) | **9771 (1.00×)** | 14889 (1.52×) | | add, complete and filter todos | 16997 (1.10×) | 20867 (1.35×) | **15503 (1.00×)** | 21196 (1.37×) | | open a todo, mark it done, then delete it | **12015 (1.00×)** | 18881 (1.57×) | 12491 (1.04×) | 18171 (1.51×) | | deep link opens a todo by id | 11791 (1.13×) | 17367 (1.67×) | **10417 (1.00×)** | 18056 (1.73×) | | settings shows the account, toggles switches and signs out | **13018 (1.00×)** | 19116 (1.47×) | 13228 (1.02×) | 18070 (1.39×) | ### Per-step latency [#per-step-latency] | Framework | Steps | Median | p95 | Max | | ------------ | ----- | ------ | ------ | ------ | | warden | 504 | 170 ms | 3.93 s | 5.00 s | | argent | 1192 | 318 ms | 1.82 s | 2.82 s | | agent-device | 632 | 479 ms | 1.62 s | 1.90 s | | tester-army | 592 | 494 ms | 3.18 s | 3.98 s | ### Untraced time [#untraced-time] | Framework | Untraced / rep | Share of test time | Largest (test) | | --------- | -------------- | ------------------ | ------------------------------------------- | | warden | 3.35 s | 3.5% | settings/account-switches-sign-out (718 ms) | Test wall time that no step and no span of the framework's own trace covers (warden: `trace.jsonl`; nested steps and spans count once). Runner overhead, hooks, sleeps and calls nobody wrapped end up here. ### Resources [#resources] | Framework | Scope | CPU-s / rep | CPU-s (whole) | CPU avg | CPU peak | RSS avg | RSS peak | | ------------ | ----- | ----------- | ------------- | ------- | -------- | -------- | -------- | | warden | host | 9.12 s | 46.25 s | 8.1% | 59.4% | 147.8 MB | 184.9 MB | | warden | sim | — | 563.94 s | 102.7% | 292.5% | 36.4 GB | 36.5 GB | | argent | host | 2.79 s | 14.69 s | 1.7% | 39.7% | 182.0 MB | 261.8 MB | | argent | sim | — | 956.48 s | 131.3% | 788.2% | 36.3 GB | 36.5 GB | | agent-device | host | 10.30 s | 73.33 s | 12.9% | 532.8% | 745.9 MB | 2.4 GB | | agent-device | sim | — | 683.96 s | 125.8% | 380.0% | 36.8 GB | 37.0 GB | | tester-army | host | 17.01 s | 107.19 s | 13.9% | 508.1% | 599.5 MB | 2.0 GB | | tester-army | sim | — | 914.59 s | 118.9% | 560.4% | 36.8 GB | 37.0 GB | Sampled every 500 ms from `ps`. `host` = the framework's own process tree; `sim` = every process of the simulator's `launchd_sim` tree, or on Android the app under test plus every framework's on-device helpers from `top` (every 2 s). RSS is a per-process sum, so shared framework pages are counted once per process: compare the numbers across frameworks, not against physical memory. `CPU-s / rep` (the CPU headline) is the CPU time of the timed run only, median over the warm reps (over every rep, cold included, when a framework has no warm rep, as with `--runs 1`), from per-process deltas of `ps time`; the other columns are whole-framework (prepare, every rep, the per-rep housekeeping and teardown), so their averages move with suite length. `CPU avg` / `CPU peak` are `ps %cpu`, a decaying average on macOS, kept for comparison with older runs. Android `sim` has no CPU-seconds (`top` only). CPU-seconds undercount: CPU a process uses after the last 500 ms tick, and processes living between ticks, are only recovered for children the bench ran itself (their rusage at exit); daemons and processes a framework spawns and never waits for are not, so frameworks that fan out to many short-lived tools read low. An exit charge is booked to its root's group, so in the per-group split the unseen CPU of a descendant of another group (a `bun test` under a `warden` root) shows under the root. procps (Linux) `ps time` has 1 s resolution. #### Host by process group [#host-by-process-group] | Framework | Group | CPU-s (whole) | RSS avg | RSS peak | | ------------ | ----------- | ------------- | -------- | -------- | | warden | bun-test | 10.27 s | 92.0 MB | 129.3 MB | | warden | warden-simd | 35.39 s | 55.7 MB | 56.2 MB | | warden | warden-cli | 0.13 s | 82.2 KB | 77.7 MB | | warden | other | 0.47 s | 21.7 KB | 20.5 MB | | argent | other | 14.69 s | 182.0 MB | 261.8 MB | | agent-device | other | 73.33 s | 745.9 MB | 2.4 GB | | tester-army | other | 107.19 s | 599.5 MB | 2.0 GB | `bun-test` the spec process, `warden-simd` the resident iOS helper, `warden-cli` the `warden` commands the bench spawned, `adb` / `emulator` the Android tools, `other` the rest (another framework's runner, `xcrun`, java). Each group's RSS average counts 0 for ticks without a process, so the groups add up to the host. ### Where the CPU goes [#where-the-cpu-goes] #### warden [#warden] | Process | Scope | CPU-s | Share | RSS avg | RSS peak | Spawns | | ---------------------------------- | ----- | ------- | ----- | -------- | -------- | ------ | | warden-simd | host | 7.03 s | 77.2% | 55.9 MB | 56.0 MB | 1 | | bun test | host | 1.15 s | 12.6% | 92.2 MB | 107.1 MB | 1 | | bun test (exit charge) | host | 0.69 s | 7.5% | 0 B | 0 B | 0 | | simctl | host | 0.24 s | 2.7% | 277.4 KB | 20.7 MB | 3 | | (PlistBuddy) | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | (bash) | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | (simctl) | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | bun cli.ts | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | iOS | sim | 75.30 s | 64.7% | 34.8 GB | 34.8 GB | 286 | | wardenexample | sim | 30.57 s | 26.3% | 374.9 MB | 432.8 MB | 8 | | SimulatorFrameworkBridge-iOS serve | sim | 6.04 s | 5.2% | 197.7 MB | 210.2 MB | 12 | | bun cli.ts | sim | 2.01 s | 1.7% | 235.0 MB | 241.7 MB | 1 | | CoreSimulatorBridge | sim | 0.94 s | 0.8% | 83.7 MB | 83.9 MB | 1 | | launchd\_sim | sim | 0.39 s | 0.3% | 52.2 MB | 52.2 MB | 1 | | xpcproxy\_sim | sim | 0.26 s | 0.2% | 390.5 KB | 11.4 MB | 8 | | mlhostd | sim | 0.24 s | 0.2% | 129.7 KB | 14.4 MB | 2 | | ps | sim | 0.15 s | 0.1% | 1.9 MB | 1.9 MB | 191 | | (launchctl) | sim | 0.14 s | 0.1% | 93.2 KB | 17.4 MB | 1 | | storekitd | sim | 0.14 s | 0.1% | 127.9 MB | 127.9 MB | 1 | | triald | sim | 0.13 s | 0.1% | 96.6 MB | 96.6 MB | 1 | | other | sim | 0.00 s | 0.0% | 443.9 MB | 443.9 MB | 6 | #### argent [#argent] | Process | Scope | CPU-s | Share | RSS avg | RSS peak | Spawns | | ------------------------- | ----- | ------- | ----- | -------- | -------- | ------ | | simulator-server | host | 2.50 s | 92.0% | 96.6 MB | 96.7 MB | 1 | | node argent | host | 0.21 s | 7.8% | 85.7 MB | 87.9 MB | 1 | | node argent (exit charge) | host | 0.01 s | 0.2% | 0 B | 0 B | 0 | | (node) | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | node tool-server.cjs | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | simctl | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | system\_profiler | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | iOS | sim | 98.77 s | 74.0% | 34.8 GB | 34.9 GB | 287 | | wardenexample | sim | 29.50 s | 22.1% | 375.5 MB | 435.1 MB | 8 | | bun cli.ts | sim | 2.78 s | 2.1% | 298.6 MB | 308.0 MB | 1 | | simctl | sim | 0.41 s | 0.3% | 363.8 KB | 20.5 MB | 7 | | CoreSimulatorBridge | sim | 0.40 s | 0.3% | 117.8 MB | 118.2 MB | 1 | | launchd\_sim | sim | 0.37 s | 0.3% | 52.2 MB | 52.2 MB | 1 | | ps | sim | 0.29 s | 0.2% | 1.9 MB | 2.0 MB | 260 | | (mlhostd) | sim | 0.26 s | 0.2% | 222.5 KB | 40.5 MB | 2 | | triald | sim | 0.20 s | 0.1% | 96.7 MB | 96.7 MB | 1 | | storekitd | sim | 0.19 s | 0.1% | 128.4 MB | 128.4 MB | 1 | | (managedappdistri) | sim | 0.14 s | 0.1% | 214.5 KB | 54.5 MB | 1 | | mlhostd | sim | 0.13 s | 0.1% | 119.3 KB | 15.7 MB | 2 | | other | sim | 0.05 s | 0.0% | 444.5 MB | 453.2 MB | 9 | #### agent-device [#agent-device] | Process | Scope | CPU-s | Share | RSS avg | RSS peak | Spawns | | ------------------------------- | ----- | ------- | ----- | -------- | -------- | ------ | | xcodebuild | host | 7.50 s | 73.3% | 218.0 MB | 218.7 MB | 1 | | bun cli.ts | host | 1.77 s | 17.3% | 268.7 MB | 273.3 MB | 1 | | snapshot-bridge | host | 0.77 s | 7.5% | 116.3 MB | 116.4 MB | 1 | | node agent-device | host | 0.15 s | 1.5% | 61.0 MB | 66.8 MB | 1 | | ps | host | 0.02 s | 0.2% | 1.9 MB | 1.9 MB | 182 | | simctl | host | 0.02 s | 0.2% | 20.4 MB | 20.4 MB | 1 | | node agent-device (exit charge) | host | 0.01 s | 0.1% | 0 B | 0 B | 0 | | (simctl) | host | 0.00 s | 0.0% | 96.9 KB | 17.2 MB | 1 | | DTServiceHub | host | 0.00 s | 0.0% | 16.7 MB | 16.7 MB | 1 | | caffeinate | host | 0.00 s | 0.0% | 7.3 MB | 7.3 MB | 1 | | (bash) | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | (clang) | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | other | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | iOS | sim | 89.31 s | 65.7% | 35.5 GB | 35.5 GB | 291 | | wardenexample | sim | 35.81 s | 26.3% | 371.9 MB | 438.3 MB | 8 | | AgentDeviceRunnerUITests-Runner | sim | 9.14 s | 6.7% | 268.2 MB | 268.2 MB | 1 | | CoreSimulatorBridge | sim | 0.51 s | 0.4% | 80.6 MB | 80.7 MB | 1 | | simctl | sim | 0.48 s | 0.4% | 676.7 KB | 20.5 MB | 7 | | launchd\_sim | sim | 0.31 s | 0.2% | 52.4 MB | 52.4 MB | 1 | | triald | sim | 0.13 s | 0.1% | 96.7 MB | 96.7 MB | 1 | | storekitd | sim | 0.13 s | 0.1% | 128.3 MB | 128.3 MB | 1 | | xpcproxy\_sim | sim | 0.09 s | 0.1% | 176.0 KB | 9.1 MB | 4 | | mlhostd | sim | 0.08 s | 0.1% | 34.2 KB | 6.1 MB | 1 | | (wardenexample) | sim | 0.00 s | 0.0% | 2.4 MB | 433.2 MB | 1 | | assetsd | sim | 0.00 s | 0.0% | 163.2 MB | 163.2 MB | 1 | | other | sim | 0.00 s | 0.0% | 274.3 MB | 274.3 MB | 4 | #### tester-army [#tester-army] | Process | Scope | CPU-s | Share | RSS avg | RSS peak | Spawns | | ------------------------------- | ----- | -------- | ----- | -------- | -------- | ------ | | xcodebuild | host | 14.26 s | 84.0% | 217.0 MB | 218.2 MB | 1 | | node entry.js | host | 1.05 s | 6.2% | 84.9 MB | 92.8 MB | 1 | | snapshot-bridge | host | 0.63 s | 3.7% | 116.2 MB | 116.3 MB | 1 | | (node) (exit charge) | host | 0.62 s | 3.7% | 0 B | 0 B | 0 | | node e2e | host | 0.35 s | 2.1% | 109.4 MB | 132.3 MB | 1 | | (node) | host | 0.03 s | 0.2% | 0 B | 0 B | 0 | | other (exit charge) | host | 0.03 s | 0.2% | 0 B | 0 B | 0 | | (ps) | host | 0.01 s | 0.1% | 27.9 KB | 1.7 MB | 5 | | simctl | host | 0.00 s | 0.0% | 20.5 MB | 20.5 MB | 1 | | DTServiceHub | host | 0.00 s | 0.0% | 16.8 MB | 16.8 MB | 1 | | (pgrep) | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | (simctl) | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | other | host | 0.00 s | 0.0% | 0 B | 0 B | 0 | | iOS | sim | 107.48 s | 60.4% | 35.1 GB | 35.2 GB | 289 | | wardenexample | sim | 49.85 s | 28.0% | 380.4 MB | 441.8 MB | 9 | | AgentDeviceRunnerUITests-Runner | sim | 15.86 s | 8.9% | 271.0 MB | 271.2 MB | 1 | | bun cli.ts | sim | 2.56 s | 1.4% | 254.0 MB | 264.4 MB | 1 | | CoreSimulatorBridge | sim | 0.57 s | 0.3% | 80.6 MB | 80.6 MB | 1 | | simctl | sim | 0.54 s | 0.3% | 534.1 KB | 20.8 MB | 8 | | launchd\_sim | sim | 0.46 s | 0.3% | 51.8 MB | 51.8 MB | 1 | | mlhostd | sim | 0.24 s | 0.1% | 135.3 KB | 14.3 MB | 3 | | triald | sim | 0.16 s | 0.1% | 96.6 MB | 96.7 MB | 1 | | storekitd | sim | 0.15 s | 0.1% | 128.2 MB | 128.2 MB | 1 | | launchctl | sim | 0.10 s | 0.1% | 36.7 KB | 9.4 MB | 1 | | ps | sim | 0.06 s | 0.0% | 1.9 MB | 1.9 MB | 263 | | other | sim | 0.07 s | 0.0% | 446.4 MB | 870.8 MB | 8 | One table per framework, from the warm-median rep's timed run (the rep whose host CPU-seconds is the lower median). Rows are processes grouped by label (top 12 per scope, the rest in `other`); `host` is the framework's own process tree, `sim` the simulator's `launchd_sim` tree (iOS only). `Share` is of the scope's CPU-seconds. `RSS` is the label's summed RSS per 500 ms tick (0 where it had no process). `Spawns` counts distinct pids a tick saw: a process that lives between two ticks is only seen through the `(exit charge)` rows (CPU of a child the bench ran, booked at its exit), so many short-lived `xcrun simctl` calls read low in `Spawns` and in CPU. ### Residue [#residue] * **warden:** none * **argent:** paths: /Users/expo/.argent (328.0 KB) * **agent-device:** driver apps: com.callstack.agentdevice.runner.uitests.xctrunner; paths: /Users/expo/.agent-device-bench/agent-device (2.9 MB) * **tester-army:** driver apps: com.callstack.agentdevice.runner.uitests.xctrunner; paths: /Users/expo/.agent-device-bench/tester-army (4.9 MB) ## Artifacts [#artifacts] Each worker uploaded its results directory as one archive. The viewable files are copied here; the archives hold everything, including the per-rep `resources.ndjson` ticks and the 60 s idle control. To re-render one: download the archive, `tar -xzf` it, then `bun run bench report Users/expo/bench-results` from the warden repo. ### Merged run [#merged-run] The four results merged into one run, `bench-2026-10-25-ios-eas`, ratios recomputed across the four: [report.html](/bench/ios-2026-10-25/merged/report.html), [report.md](/bench/ios-2026-10-25/merged/report.md), [bench-results.json](/bench/ios-2026-10-25/merged/bench-results.json), [summary.json](/bench/ios-2026-10-25/merged/summary.json) (the comparison page's dataset). ### warden [#warden-1] Worker job `bench_warden`, bun 1.3.14 + warden-e2e (this commit). [Download the whole artifact](/bench/ios-2026-10-25/warden.tar.gz) (the EAS `eas-ios-warden` archive, unpacks to `Users/expo/bench-results`). * [report.html](/bench/ios-2026-10-25/warden/report.html) (self-contained, with charts) and [report.md](/bench/ios-2026-10-25/warden/report.md) * [bench-results.json](/bench/ios-2026-10-25/warden/bench-results.json) (schema v1, every rep, every test, every step, the 500 ms resource ticks per rep are in the archive) and [summary.json](/bench/ios-2026-10-25/warden/summary.json) * [bench-run.log](/bench/ios-2026-10-25/warden/bench-run.log) (the harness), [fw-install.log](/bench/ios-2026-10-25/warden/fw-install.log), [simd-build.log](/bench/ios-2026-10-25/warden/simd-build.log), [host-facts.txt](/bench/ios-2026-10-25/warden/host-facts.txt) * Framework logs: [bench.log](/bench/ios-2026-10-25/warden/logs/bench.log), [warden.prepare.log](/bench/ios-2026-10-25/warden/logs/warden.prepare.log), [warden.rep-1.log](/bench/ios-2026-10-25/warden/logs/warden.rep-1.log), [warden.rep-2.log](/bench/ios-2026-10-25/warden/logs/warden.rep-2.log), [warden.rep-3.log](/bench/ios-2026-10-25/warden/logs/warden.rep-3.log), [warden.rep-4.log](/bench/ios-2026-10-25/warden/logs/warden.rep-4.log), [warden.rep-5.log](/bench/ios-2026-10-25/warden/logs/warden.rep-5.log), [warden.teardown.log](/bench/ios-2026-10-25/warden/logs/warden.teardown.log) * The simulator as the job left it, after teardown: The warden worker's simulator at exit ### Argent [#argent-1] Worker job `bench_argent`, @swmansion/argent 0.30.1. [Download the whole artifact](/bench/ios-2026-10-25/argent.tar.gz) (the EAS `eas-ios-argent` archive, unpacks to `Users/expo/bench-results`). * [report.html](/bench/ios-2026-10-25/argent/report.html) (self-contained, with charts) and [report.md](/bench/ios-2026-10-25/argent/report.md) * [bench-results.json](/bench/ios-2026-10-25/argent/bench-results.json) (schema v1, every rep, every test, every step, the 500 ms resource ticks per rep are in the archive) and [summary.json](/bench/ios-2026-10-25/argent/summary.json) * [bench-run.log](/bench/ios-2026-10-25/argent/bench-run.log) (the harness), [fw-install.log](/bench/ios-2026-10-25/argent/fw-install.log), [simd-build.log](/bench/ios-2026-10-25/argent/simd-build.log), [host-facts.txt](/bench/ios-2026-10-25/argent/host-facts.txt) * Framework logs: [bench.log](/bench/ios-2026-10-25/argent/logs/bench.log) * The simulator as the job left it, after teardown: The Argent worker's simulator at exit ### agent-device [#agent-device-1] Worker job `bench_agent_device`, agent-device 0.21.31. [Download the whole artifact](/bench/ios-2026-10-25/agent-device.tar.gz) (the EAS `eas-ios-agent-device` archive, unpacks to `Users/expo/bench-results`). * [report.html](/bench/ios-2026-10-25/agent-device/report.html) (self-contained, with charts) and [report.md](/bench/ios-2026-10-25/agent-device/report.md) * [bench-results.json](/bench/ios-2026-10-25/agent-device/bench-results.json) (schema v1, every rep, every test, every step, the 500 ms resource ticks per rep are in the archive) and [summary.json](/bench/ios-2026-10-25/agent-device/summary.json) * [bench-run.log](/bench/ios-2026-10-25/agent-device/bench-run.log) (the harness), [fw-install.log](/bench/ios-2026-10-25/agent-device/fw-install.log), [simd-build.log](/bench/ios-2026-10-25/agent-device/simd-build.log), [host-facts.txt](/bench/ios-2026-10-25/agent-device/host-facts.txt) * Framework logs: [agent-device.prepare.log](/bench/ios-2026-10-25/agent-device/logs/agent-device.prepare.log), [bench.log](/bench/ios-2026-10-25/agent-device/logs/bench.log) * The simulator as the job left it, after teardown: The agent-device worker's simulator at exit ### tester.army [#testerarmy] Worker job `bench_tester_army`, e2e 0.20.0, bundling agent-device 0.21.25. [Download the whole artifact](/bench/ios-2026-10-25/tester-army.tar.gz) (the EAS `eas-ios-tester-army` archive, unpacks to `Users/expo/bench-results`). * [report.html](/bench/ios-2026-10-25/tester-army/report.html) (self-contained, with charts) and [report.md](/bench/ios-2026-10-25/tester-army/report.md) * [bench-results.json](/bench/ios-2026-10-25/tester-army/bench-results.json) (schema v1, every rep, every test, every step, the 500 ms resource ticks per rep are in the archive) and [summary.json](/bench/ios-2026-10-25/tester-army/summary.json) * [bench-run.log](/bench/ios-2026-10-25/tester-army/bench-run.log) (the harness), [fw-install.log](/bench/ios-2026-10-25/tester-army/fw-install.log), [simd-build.log](/bench/ios-2026-10-25/tester-army/simd-build.log), [host-facts.txt](/bench/ios-2026-10-25/tester-army/host-facts.txt) * Framework logs: [bench.log](/bench/ios-2026-10-25/tester-army/logs/bench.log), [tester-army.rep-1.log](/bench/ios-2026-10-25/tester-army/logs/tester-army.rep-1.log), [tester-army.rep-2.log](/bench/ios-2026-10-25/tester-army/logs/tester-army.rep-2.log), [tester-army.rep-3.log](/bench/ios-2026-10-25/tester-army/logs/tester-army.rep-3.log), [tester-army.rep-4.log](/bench/ios-2026-10-25/tester-army/logs/tester-army.rep-4.log), [tester-army.rep-5.log](/bench/ios-2026-10-25/tester-army/logs/tester-army.rep-5.log), [tester-army.teardown.log](/bench/ios-2026-10-25/tester-army/logs/tester-army.teardown.log) * The simulator as the job left it, after teardown: The tester.army worker's simulator at exit ## Follow-up: warden after the helper fix [#follow-up-warden-after-the-helper-fix] The profile in [`docs/bench/cpu-profile-2026-10-25.md`](https://github.com/delacournz/warden/blob/main/docs/bench/cpu-profile-2026-10-25.md) put 74 % of `warden-simd`'s CPU in idb's generic `Codable` decode of every accessibility frame. The helper now links idb with a single-pass frame decoder (fork commit [`1768a0d`](https://github.com/UrbanChrisy/idb/commit/1768a0d1006ff528c2fd78f335d9e3ee2f01596d) on v1.7.2, library package [`idb-v1.7.2-warden.1`](https://github.com/delacournz/warden/releases/tag/idb-v1.7.2-warden.1)). Same workflow, same worker type and simulator, warden only, 5 runs: run [`01a1739d`](https://expo.dev/accounts/delacour/projects/warden/workflows/01a1739d-2f5f-70bb-9c22-715197980e70) at commit `a87b04cc`. The bench app was rebuilt by EAS for this run (sha256 `c283eae41c7a`, same source as the run above), so the comparison page keeps the four-framework run above as its dataset and this run stands beside it. | warden, per rep (warm) | Run above (`01a172f5`) | After (`01a1739d`) | Delta | | ---------------------- | ---------------------- | ------------------ | --------------- | | Host CPU-seconds | 9.12 s | 3.79 s | -5.33 s (-58 %) | | of which `warden-simd` | 7.03 s | 1.83 s | -5.20 s (-74 %) | | Host RSS peak | 184.9 MB | 177.6 MB | -7.3 MB | | Suite median | 97.04 s | 91.89 s | -5.15 s (-5 %) | | Pass rate / flakes | 100 % / 0 | 100 % / 0 | | The simulator's own CPU over the whole job (every process under `launchd_sim`, prepare and all five reps) read higher on this worker (821 CPU-s against 564 CPU-s) while both simulators idled alike before the run (idle control in the artifacts); the host is now out of the way, so the in-simulator readers are the next thing to profile. Worker job `bench_warden`. [Download the whole artifact](/bench/ios-2026-10-25/warden-after/warden.tar.gz) (the EAS `eas-ios-warden` archive, unpacks to `Users/expo/bench-results`). * [report.html](/bench/ios-2026-10-25/warden-after/report.html) and [report.md](/bench/ios-2026-10-25/warden-after/report.md) * [bench-results.json](/bench/ios-2026-10-25/warden-after/bench-results.json) and [summary.json](/bench/ios-2026-10-25/warden-after/summary.json) * [host-facts.txt](/bench/ios-2026-10-25/warden-after/host-facts.txt), harness and framework logs under [`logs/`](/bench/ios-2026-10-25/warden-after/logs/bench.log) * The simulator as the job left it, after teardown: The warden worker's simulator at exit, after the helper fix